THE NIGERIA TELECOMMUNICATIONS GIANTS

AND THEIR SOCIAL MEDIA HEALTH

Derek Osawaguan Ahamioje

Derek Ahamioje 
ahamiojederek@gmail.com 
1 
Table of Contents 
INTRODUCTION .................................................................................................................................. 3 
1  PREAMBLE .............................................................................................................................. 3 
1.1  TELECOMMUNICATION IN NIGERIA ....................................................................................... 3 
1.2  BUSINESS ANALYTICS ............................................................................................................. 4 
1.3  AIMS AND OBJECTIVES ........................................................................................................... 5 
2.1  DATA ....................................................................................................................................... 6 
2.2  ANALYSIS AND VISUALISATION ............................................................................................. 6 
2.3  DATA HARVESTING ................................................................................................................. 6 
2.3.1  TWITTER .......................................................................................................................... 7 
2.3.2  TWINT API ....................................................................................................................... 7 
2.4  DATA ANALYSIS AND VISUALIZATION ................................................................................... 8 
2.4.1  Tableau ........................................................................................................................... 8 
2.4.2  Python ............................................................................................................................ 8 
RESULTS AND DISCUSSIONS ................................................................................................................. 10 
CONCLUSIONS, PROJECT LIMITATIONS, AND RECOMMENDATIONS ................................................. 25 
4.1  Conclusion ............................................................................................................................... 25 
4.2  LIMITATIONS ......................................................................................................................... 25 
4.2  RECOMMENDATIONS ........................................................................................................... 25 
REFERENCES ................................................................................................................................... 26 
 
 
 
 
 
 
 
 
 
 
 
 

Derek Ahamioje

ahamiojederek@gmail.com

Abstract

Purpose: There is a strong correlation between the perception of customers toward a business

service and the level of patronage towards the business. In a situation where there is a positive

perception, the tendency to subscribe to the service will be high. Most businesses have

deployed social media as a medium for selling their products and obtaining feedback from both

their customers and prospective customers. Therefore, this study uses the social media analysis

of four major telecommunications providers in Nigeria (MTN Nigeria, Globacom, Airtel, and

9mobile) to extract useful sentiments.

Design: 5000 tweets from the Twitter handle of each of the service provider was scraped and

visualized using Tableau, a visualisation tool, and python visualization with latent Dirichlet

allocation technique.

Findings: It was observed that responses from customers and potential customers were higher

when providers made offers. Providers like MTN with larger followers enjoyed pleasurable

sentiments when compared with others like 9moblie and Airtel with very low followership,

which had dissatisfactory sentiments.

Business Implications: The implications for these service providers are that this analysis

presents before them, the unscrutinized, uncontrolled, and cheap feedback system for

identifying key areas of customer satisfaction.

Derek Ahamioje

ahamiojederek@gmail.com

INTRODUCTION

1 PREAMBLE

This research is aimed at gaining a deeper understanding of the perception of

telecommunications subscribers in Nigeria towards their chosen telecommunications

service providers. It contains background on the telecommunication industry in

Nigeria, the major service providers and the objectives of this study, design method,

analysis, results evaluation, and conclusion from the study.

1.1 TELECOMMUNICATION IN NIGERIA

The Telecommunications industry in Nigeria has over the decades experienced a huge

increase in the number of subscribers across the six geopolitical zones of the country.

This advancement no doubt has contributed in no small way to improving the quality

of life of these subscribers, from increasing communication links among families to

enhancing customer reach (CR) for businesses and translating to an increase in

business profit. However, in Tekanyi et al., 2019 while it was acknowledged that the

Global System for Mobile (GSM)

communications technology has become one of the fastest growing technologies, it

also affirms that the telecommunications technology is one of the most challenging

technologies. These challenges were assessed for efficiency in terms of the quality of

services delivered on the basis of The Call Setup Failure Rate (CSFR) and Hand-Over

Failure Rate (HOFR). From the business perspective of the Telecommunication

industries, a key predictive indicator for their profit will depend largely upon the level

of satisfaction that their customers enjoy for subscribing to the services offered to

them. However, according to Tekanyi et al., 2019, maximum customer satisfaction has

not been the case over the years resulting from the increase in Hand-over Failure Rate

and poor network availabilities due to an increased Call Setup Failure Rate which has

become a great source of concern to providers, users, and researchers.

Ononiwu et al., 2016 while reviewing four major GSM providers in Nigeria (Globacom,

Airtel, 9mobile, and MTN) opined that by continuously tracking, evaluating, and

enhancing the radio network performance (RNP) to deliver higher subscriber quality,

Mobile Network Operators (MNO) may dramatically increase their profitability growth

rates and financial stability. The research feedback conducted in the south-east part

of the country using the RNP indicator showed that these network providers failed to

comply with the Stipulated Performance Threshold (SPT) as recommended by the

Derek Ahamioje

ahamiojederek@gmail.com

Nigerian Communications Commission (NCC), a regulating body for communication

Networks in Nigeria.

Abayomi (2011), had conducted similar research in the southwestern part of Nigeria

by performing T-test, F-test, [at 95% confidence limit], Cross-tabulation [using the ‘Eta’

Directional measure], and statistical charts on a set of data extracted from a user

survey conducted in the region. This was done to assess subscribers’ satisfaction with

their chosen GSM service providers.

According to Folorunsho and Nkemdilim (2020), the GSM market in Nigeria is one of

the fastest growing in the world and contributes around 10% of the country's GDP.

According to Oghojafor et al., (2014), the sector has fallen behind the other industries

in terms of consumer satisfaction and loyalty. Sanusi and Oloyede, (2014) while

analysing the telecommunications sector in Nigeria, had identified the biggest players

in the industry otherwise code-named “the big four” as - MTN Nigeria, Glo Mobile

(Globacom), 9Mobile, and Airtel Nigeria. The performance evaluation by the Nigerian

Communications Commission (NCC) in terms of market shares is shown in Figure 1,

where MTN Nigeria ranked highest in market shares with an aggregate of 38% even

though it was a record drop from 39.61 percent of the market share the previous year

(Folorunsho and Nkemdillim, 2020). Glo and Airtel at equal shares of 28%. While

9mobile had a 9% market share.

However, beyond these measures used in assessing the service providers, recent

development has seen the use of machine learning algorithm to extract the true intent

of a writer about a specific subject through their uncontrolled written text, which is

otherwise described as sentiment analysis; subjectivity analysis, opinion mining and

appraisal extraction (Mejova, 2009). This can better give an insight into the following

analytical question.

1. Are the subscribers in Nigeria satisfied with their network providers?

2. What are they saying about the services they receive?

1.2 BUSINESS ANALYTICS

Gaining new insights into business is important for a business desiring to improve its

performance in several ways such as; quality service delivery, growing the business

profit margin, identifying new markets or potential markets, minimizing loss, etc.

Derek Ahamioje

ahamiojederek@gmail.com

Therefore, as organizations are working hard to have a competitive advantage over

their competitors, business analytics has become increasingly important in achieving

the competitive advantage (Mikalef et al., 2019), to which several studies have

underscored the positive implications business analytics has had on business output

(Ashrafi et al., 2019)).

1.3 AIMS AND OBJECTIVES

This study aims to extract meaningful sentimental feedback from the service

subscribers through the social media handle of the respective network providers and

use such feedback to propose solutions to areas of improvement for the service

provider. To execute this aim, the following objectives have been put in place;

1. Identify the effective Twitter handle for Nigeria’s top four telecoms provider.

2. Extract relevant data including feedback from the social media hand using a

suitable API

3. Visualise the extracted data using Tableau and Python

4. Identify areas of improvement for each network provider for an enhanced

customer base

Figure 1: Market Share of Mobile GSM Operators in Nigeria – December 2021

(NCC. 2022)

Derek Ahamioje

ahamiojederek@gmail.com

RESEARCH DESIGN

The research work was done using the following grouping: Data and Analysis.

2.1 DATA

The aim of the study is to analyse the social media feeds of the selected telecoms

providers in Nigeria, therefore in the task of obtaining the dataset, the choice of social

media handle that will best provide these data was drawn from some important

characteristics' questions, such as:

I. How many social media handles are these telecoms operating?

II. How active are they in the media?

III. Which medium provides sufficient data for this analysis?

The data under investigation are both historical datasets and real-time. The Historical

dataset is the obtained data covering the dates preceding the date the data was

harvested, while the real-time data were the data harvested at the exact time it was

posted on the page. The two focuses in terms of the type of data in this study are

textual data and numerical data and combining both to make a useful conclusion.

2.2 ANALYSIS AND VISUALISATION

The next phase of the study is to carry out some analysis of the data obtained. This is

the data analytics stage, which is the process of examining the data to observe

possible trends and therefore make informed conclusions from the observations.

2.3 DATA HARVESTING

According to Batrinca, and Treleaven, (2015), Social media data is often either

accessible through a simple and straightforward general method or may require the

researcher to programme their analytics in a language like MATLAB, Java, or Python.

The dataset used for the analysis of the social media feeds of the chosen

telecommunications companies (MTN Nigeria, Globacom Nigeria, Airtel Nigeria, and

9mobile Nigeria), was obtained from Twitter by a technique known as scraping and

using the Python language.

Derek Ahamioje

ahamiojederek@gmail.com

2.3.1 TWITTER

Twitter a foremost social media provider was the source of the data used in the study.

Unlike some providers with unrestricted access to data, Twitter allows only restricted

access to data in its platform, therefore, denying access to the whole data. It, however,

grants limited access right for research purposes through its approved API (application

programming interface), which must be obtained through the predefined process of

application and approval for a developer account. In this study, the effort to secure a

developer account was difficult based on the fact that the approval was being delayed

while the timeline for the study was running out. However, twint was identified as an

API that can scrape data.

2.3.2 TWINT API

Twint is an advanced tool for data scraping without the need for Twitter API, and it is

written in Python programming language. Twint is an awesome API when it comes to

obtaining tweets, but it also has its limitations when it comes to interacting with Twitter

(such as making a post or sending direct messages). In this study, our focus is

obtaining tweets therefore, twint is just sufficient for use. Figure 1 shows the features

of the data scraped using twint.

Figure 1: Features of data obtained from Twitter using twint API

For an effective analysis, a total of five thousand (5,000) tweets were scraped with 38

features each as shown in Figure 1 above, for MTN, Globacom, 9mobile, and Airtel

respectively using python language. The key features that address the aim were

identified for analysis and visualisation

Derek Ahamioje

ahamiojederek@gmail.com

2.4 DATA ANALYSIS AND VISUALIZATION

The data scraped from the Twitter handle of the four telecommunication providers in

Nigeria were subjected to a standard algorithm of analysis and visualisation using a

combination of two visualisation tools; Tableau and Python.

2.4.1 Tableau

Tableau, a data visualisation tool was used for the pictorial and visual representation

of the data to observe the interactions between chosen features in this analysis. The

choice of Tableau for visualisation was made because of its simplicity, ease of use for

new users including those with minimum experience, and the amazing visualisation it

can generate. Tableau was used on the numeric data to analyse interactions between

the providers and their subscribers, using features such as; the number of likes,

number of replies, and number of retweets.

Figure 2: Sample display of Tableau desktop (Source: Patel, 2021)

2.4.2 Python

Python language, a fifth-generation programming language was used to build the

Latent Dirichlet Allocation (LDA) model; a generative probabilistic model for collections

of discrete data such as text corpora (Blei, et al., 2003). The model was used to create

visualisations based on topic relevance using the gensim python library thereby

presenting subscribers’ opinions in a visual form.

Derek Ahamioje

ahamiojederek@gmail.com

The choice of using python for this purpose was because it is a familiar tool for textual

data analysis, and its use by researchers in analysing tweets from Twitter (Negara, et

al., 2019).

Derek Ahamioje

ahamiojederek@gmail.com

RESULTS AND DISCUSSIONS

Results obtained from the study show that a total of 5000 tweets were scraped with

twint API for GloWorld, 9mobile, and Airtel while for MTN, 5001 tweets were scraped

using the same size limits of 5000. The tweets scraped covered different periods,

based on the number of tweets from each handle. MTN Nigeria had the tweet spread

from the 15

of January, 2019 to the 31

of March, 2023, and Globacom had its entire

number of tweets from the 24

of February 2023 to the 31

of March 2023. However,

9mobile and Airtel tweets spread from the 20

of February, 2023 to the 31

of March

and the 25

of March to the 31

of March, 2023 respectively. Figure 1 shows the tweet

count for each network provider. Preliminary assessment shows that the average

tweets per day from each network provider based on the total number of tweets

scraped and the time stamp for each tweet indicates that MTN Nigeria, 9mobile,

Globacom, and Airtel, had average tweets per day of 4, 125,139 and 714 tweets per

day, which already, raises interest as to how well will each provider perform based on

the volume of their tweets and also we can possibly draw further inference from the

content of the tweets based on the statistics shown in Table 1, where despite having

4 tweets per day, MTN Nigeria attracted a total of 334,161 likes and 120,512 replies,

while Airtel with the highest number of tweets per day recorded 236 likes and 190

replies in all.

Table 1: Preliminary results from scraped tweets for all networks

No. of

tweets

No. of

days

Average

tweets per day

Total No. of

Likes

Total No. of

Replies

MTN

Nigeria

5001

1171

334,161

120,512

GloWorld

5000

139

4,795

3247

9mobile

5000

125

Airtel

5000

714

236

1903

Derek Ahamioje

ahamiojederek@gmail.com

Figure 1: Tweet count for each network provider

Results obtained with the Tableau visualisation tool highlighted tweets in the order of

the sizes of the number of likes per tweet and number of replies per tweet, which

suggests that tweets with more likes are satisfactory to the customers more than those

with fewer likes, while those with more replies are more interactive than those with

fewer replies. On this basis, Figure 3, is a dashboard view showing the growth of the

likes per tweet that each provider attracted and the respective dates each provider

attracted the highest and lowers numbers of likes. MTN Nigeria had the top three

highest likes for tweets made in February 2022, November 2021, and January 2023

with 33,387, 29,450, and 19,315 likes respectively. Globacom had 1,099, 506, and

475 likes on 8

March 2023, 26

February 2023, and 17

March 2023 respectively.

9Mobile also had 64, 63, 62 likes on February 28

, March 24

, 2023, and March 22nd,

2023 respectively.

Derek Ahamioje

ahamiojederek@gmail.com

Figure 3: Growth of the number of likes per tweet for network providers.

However, a comparison of the number of likes with the number of replies shows that

tweets with the highest number of likes may not necessarily translate to having the

highest number of replies as shown in Figure 4. In the case of Globacom, while the

number of tweets with the highest likes is tweets of 8

March 2023, 26

February

2023, and 17

March 2023, those with the highest replies are March 12

, March 28h,

and March 15

. The same applies to Airtel, on one occasion with MTN. Observations

with 9mobile were different because the number of replies was observed to be higher

than the number of likes and in those cases, the number of likes was high where the

number of replies was also high.

Derek Ahamioje

ahamiojederek@gmail.com

Figure 4: Dashboard showing a compare between the number of likes and with

number of replies.

Furthermore, a look at the first three highest likes and first three highest replies show

that for MTN as shown in Figures 5a, b, and c, The tweet with the highest likes of

29,038 and 6547 replies was a tweet announcing a comeback in 2022 which suggests

the excitement of subscribers with the return. While the second highest (figure 5b) was

a message of inspiration at the beginning of the year, it attracted 10,827 likes and 640

replies and the third was an advert presented in the form of self-identity, which can

also be perceived as a motivation. (Figure 5c)

Derek Ahamioje

ahamiojederek@gmail.com

Figure 5a: MTN tweet with the highest number of likes

Figure 5b: MTN tweet with the second-highest number of likes

Derek Ahamioje

ahamiojederek@gmail.com

Figure 5c: MTN tweet with the third-highest number of likes

The top three highest number of likes for Globacom is shown in Figures 6a, 6b, and

6c respectively. The tweet with the highest likes was offering discounts and it

attracted a total of 630 likes and 33 replies (figure 6a), the tweet with the second

highest also offered discounts (figure 6b), while the third highest likes addresses

women empowerment on international women’s day(Figure 6c)

Figure 6a: Globacom tweet with the highest number of likes

Derek Ahamioje

ahamiojederek@gmail.com

Figure 6b: Globacom tweet with the second-highest number of likes

Figure 6c: Globacom tweet with the third highest number of likes.

The feedback from 9mobile Nigeria reveals that the tweets with the first two highest

likes were apologies. While the highest (figure 7a) was apologizing for barring a

subscriber’s line, the second was apologising for the inability to provide the

subscribers’ requested information. However, just like Globacom, the third highest

was also eulogizing women on international women’s Day.

Derek Ahamioje

ahamiojederek@gmail.com

Figure 7a: 9Moblie tweet with the highest number of likes

Figure 7b: 9Mobile tweet with the second highest number of tweets

Derek Ahamioje

ahamiojederek@gmail.com

Figure 7c: 9Mobile tweet with the third highest number of likes

However, in the case of Airtel, the number of likes and replies per tweet was very low

with the highest out of a total of 5000 tweets being 6 suggesting that Airtel’s presence

on Twitter is very low and insufficient for its assessment. Appendices 1, 2, and 3 are

visuals of these statistics.

Sentiments analysis on the replies to tweets reveals that for MTN, the first three tweets

with the highest number of replies had an engaging approach to the tweets. While the

tweet with the highest replies presented a reward for people who will post their phone

numbers and tag their friends. See Figure 8. The same approach was used for

subscribers in the second and the third highest as shown in Figures 8b and 8c

respectively. Another notable observation from these tweets shows that the number

of replies was more than the number of likes, indicating that people were more

interested in the reward than in indicating their likeness.

It was different for GloWorld and 9mobile where their tweets were in the form of posing

fun game questions, which also, attracted more replies than likes in each post, but in

much smaller counts respectively when compared with the total counts from MTN.

Derek Ahamioje

ahamiojederek@gmail.com

Figure 8a: MTN tweet with the highest number of replies

Figure 8b: MTN tweets with 2

highest number of replies

Derek Ahamioje

ahamiojederek@gmail.com

Figure 8c: MTN tweet with 3

highest number of replies.

The topic modelling analysis performed with python language identified keywords in

the collection of tweets and using a machine learning algorithm classified keywords

into relevant topics based on semantic similarity. Figures 9a, 9b, 9c, and 10a, 10b,

10c, shows the estimated term frequency of keywords in the first three topics in MTN,

and Globacom.

Figure 9a: Terms relevance for MTN topic 1

Derek Ahamioje

ahamiojederek@gmail.com

Figure 9b: Terms relevance for MTN topic 2

Figure 9c: Terms relevance for MTN topic 3

Derek Ahamioje

ahamiojederek@gmail.com

Figure 10a: Relevant terms for GloWorld Topic 1

Figure 10b: Relevant terms for GloWorld Topic 2

Derek Ahamioje

ahamiojederek@gmail.com

Figure 10a: Relevant terms for Globacom Topic 3

A word cloud obtained from these selected topics also agrees with the result obtained

from the bubble plot in Tableau where the first three tweets with the highest replies

share the same keywords as the first three topics in the word cloud as shown in Figure

11 for instance, where the word cloud is screaming “mtn #socialmediaspree”, “Https”,

“tco” etc. On the other hand, Figure 12 shows the word cloud for Globacom screaming

more of “issue”, “sorry” , “assistance”, apologies, etc., which suggests that the

subscribers had a lot of issues using the network within the period under review.

Derek Ahamioje

ahamiojederek@gmail.com

Figure 11: word cloud for MTN’s first four topics

Figure 12: GloWorld word cloud for the first four dominant topics

Derek Ahamioje

ahamiojederek@gmail.com

CONCLUSIONS, PROJECT LIMITATIONS, AND RECOMMENDATIONS

4.1 Conclusion

Guided by the objective of this study, Twitter was identified as a social media platform

where suitable data can be extracted for the different telecom organisations. The use

of twint API, an open-source API, was able to scrape sufficient data (5000 tweets in

less than a minute) from Twitter without using the regulated Twitter API.

The result obtained from this study and, based on the analysis conducted on the

scraped data with python and Tableau, it can be concluded, that of the four network

service providers, MTN Nigeria stands out as the most attractive and satisfactory

service provider to its subscribers whom they have, regularly involved in their subject

of the day, through their juicy, on-the-go freebies and other offers on their Twitter

handle. The study further has shown that the quality of tweets is way more important

than the number of tweets, as seen in the case of MTN has an average tweet of 4 per

day and yet can attract a total of 334,161 likes and 120,512 replies, while the likes of

Airtel tweeted 5000 tweets in 7 days and yet could only attract a total of 234 likes or

1903 replies.

Furthermore, to optimize traffic to the media hand, telecom providers that are posing

brain teaser questions have advantages over those that are not using either brain

teaser questions or offering freebies.

4.2 LIMITATIONS

This study, however, experienced some limitations. The major limitation was.

Delays in securing a Twitter developer account. The implication of this is that the data

scraped with twint, did not display some features like geographical locations which

would have helped in identifying location-based sentiments.

4.2 RECOMMENDATIONS

The study has however established among others that the quality of tweets can attract

huge traffic to the providers’ page, thereby creating room for increased daily

advertisement of their products and services but did not provide answers to how it has

contributed to the company’s financial status and hence would recommend this aspect

for further study.

Derek Ahamioje

ahamiojederek@gmail.com

REFERENCES

Abayomi, A.S., 2011. GSM Marketing Service Providers Operations and Customers

Satisfaction in Nigeria: An Empirical Investigation. Acta Universitatis Danubius.

OEconomica, 7(3), pp.77-103.

Ashrafi, A., Ravasan, A. Z., Trkman, P., and Afshari, S. (2019). The role of business

analytics capabilities in bolstering firms’ agility and performance. International

Journal of Information Management, 47, 1-15.

Batrinca, B. and Treleaven, P.C., 2015. Social media analytics: a survey of techniques,

tools, and platforms. Ai & Society, 30, pp.89-116.

Blei, D.M., Ng, A.Y. and Jordan, M.I., 2003. Latent Dirichlet allocation, journal of

machine learning research 3 (Jan).

Folorunsho Ajeyalemi, O. and Nkemdillim Dixon-Ogbechie, B., 2020. AN

ASSESSMENT OF CUSTOMER PERCEPTION OF THE BRAND

PERSONALITY OF GLOBACOM, NIGERIA. Internext: Revista Electrônica de

Negócios Internacionais da ESPM, 15(3).

Mejova, Y., 2009. Sentiment analysis: An overview. University of Iowa, Computer

Science Department. Pp. 3-6

Mikalef, P., Pappas, I., Krogstie, J. and Pavlou, P., 2019. Big data and business

analytics: A research agenda for realizing business value.

NCC, 2022. 2021 SUBSCRIBER/NETWORK DATA ANNUAL REPORT, Viewed 3

April 2023. https://www.ncc.gov.ng/docman-main/industry-statistics/policies-

reports/1075-2021-year-end-subscriber-network-data-report-1/file

Negara, E.S., Triadi, D. and Andryani, R., 2019, October. Topic modeling Twitter data

with latent Dirichlet allocation method. In 2019 International Conference on

Electrical Engineering and Computer Science (ICECOS) (pp. 386-390). IEEE.

Oghojafor, B.E.A., Ladipo, K.A.P., Ighomereho, O.S. and Odunewu, A.V., 2014.

Determinants of customer satisfaction and loyalty in the Nigerian

telecommunications industry. British Journal of Marketing Studies, 2(5), pp.67-

83.

Ononiwu, G., Akinwole, B.O.H., Agubor, C. and Onojo, J., 2016. Performance

Evaluation of Major Mobile network operators in Owerri Metropolis of Nigeria.

International Association of Scientific Innovation and Research, 16(304).

Patel, A., 2021. Data Visualization Using Tableau.

Derek Ahamioje

ahamiojederek@gmail.com

Sanusi, B.O. and Oloyede, D.B., 2014. Audience perception of the role of advertising

on product positioning and brand loyalty in the Nigerian GSM Market. Kuwait

Chapter of the Arabian Journal of Business and Management Review, 3(6),

p.234.

Tekanyi, A.M.S., Abdulkareem, H.A. and Muhammad, Z.Z., 2019. Analysis of GSM

network quality of service using call setup failure rate and handover failure rate

indices. Telecommunications and radio engineering, 78(16).